2.2 Case Study: Biometric screenings results analysis
2.2.2 Objective
2.2.4.1 Descriptive Statistics
In this section, the primary data collected through biometric screenings are summarized and displayed in tables and graphics to obtain an overview of the main findings regarding demographic information and occupational problems.
Table 4 outlines the results obtained during the biometric screening event. It enumerates frequencies, means, standard deviation, and percentages in terms of participants' gender, age, occupation, height, weight, BMI, alcohol, drug consumption, the results of the blood and urine tests, and the physical examination.
Characteristic Value
Gender
Female count (%) 161 (39,4)
Male count (%) 248 (60,6)
Age
15- 41 years count (%) 240 (58,7)
Greater than 41 years count (%) 169 (41,3)
Occupation
Physical work count (%) 48 (11,7)
Mental work count (%) 361 (88,3)
Mean Height (Standard Deviation) 1,63 (0,08) Mean Height Male (Standard Deviation) 1,67 (0,71) Mean Height Female (Standard Deviation) 1,58 (0,07) Mean Weight (Standard Deviation) 71,45 (12,98) Mean Weight Male (Standard Deviation) 75,73 (11,86) Mean Weight Female (Standard Deviation) 64,86 (11,85) Mean BMI (Standard Deviation) 26,61 (3,95) Mean BMI Male (Standard Deviation) 27,12 (3,75)
52
Mean BMI Female (Standard Deviation) 25,83 (4,15) BMI Classification
Normal % 38,9
Overweight % 44
Obesity % 16,4
Morbid Obesity % 0,7
Abnormal Results Laboratory Exams
No Problems % 24,2
1 Problem % 24
2 Problems % 24,2
3 Problems % 14,9
4 Problems % 5,4
5 Problems % 5,1
6 Problems % 1,5
7 Problems % 0,5
8 Problems % 0,2
Clinical Problems
Sedentarism % 57
Polycythemia % 56,5
Hypercholesterolemia % 46,2
Hypertriglyceridemia % 36,7
Hyperuricemia % 10,5
Fatty liver % 9,1
Hypertension % 7,6
Anemia % 5,1
Urinary tract infection % 3,4
Hyperglycemia % 2,2
Problems detected during the Physical Exam
No Problems % 50,4
1 Problem % 34,2
2 Problems % 8,1
3 Problems % 3,9
53
4 Problems % 2,2
5 Problems % 1
8 Problems % 0,2
Pathologies
Metabolic % 78
Hematologic % 55, 3
Ophthalmological % 38,6
Musculoskeletal % 19,1
Urinary-genital % 3,9
Vascular % 2,7
Gastrointestinal % 1,5
Alcohol Consumption
Yes count (%) 34 (8,3)
No count (%) 375 (91,7)
Tobacco Consumption
Yes count (%) 13 (3,2)
No count (%) 396 (96,8)
Occupational Diagnosis
Fit count (%) 341 (83,4)
Fit with limitations count (%) 68 (16,6)
Table 4: Summary of the Biometric Screening Results
Regarding the participants' characteristics, males constitute a larger percentage of employees compared with females. Concerning age group, participants in the range of 15 to 41 years represent a higher percentage than their colleagues over 41 years old. A relatively small number of participants (12%) perform physical work than mental work (88%). This matter is explained by the fact that the study was executed in a university where office work is prevalent for professors, assistants, and office managers.
Concerning the measured biometric traits such as height and weight, the mean height values for males and females are within the normal ranges for Ecuadorians, which are 1.67 and 1.54 meters, respectively, according to WHO [129]. There are extreme values
54
to pay attention to obesity and diseases; this is reflected in the high standard deviation values, representing variation in the mean weight group.
Therefore, the mean BMI values for females (25.83) and males (27.11) fall into the range of pre obesity, according to WHO (25-29.9), which shows that the workforce is slightly overweight. The standard deviation values are high, which is essential to pay attention to employees under extreme BMI classifications, such as for obesity and morbid obesity[116]. Besides, the overall BMI classification shows a considerable percentage (60%) of overweight or obese employees, and nearly 40% have a normal BMI.
The laboratory exam results show that nearly 25 % are within the normal parameters regarding the blood and urine samples, while almost 50% of the participants present 1 or 2 abnormal results. In contrast to 0.7%, that has 7 or 8 problems of the 11 analyzed results.
The physical exams indicated that 50% of the employees do not have physical issues, and just 1% have 5 or 8 problems over the 18 problems presented in the whole university community.
In addition, the occupational doctor diagnosed clinical problems and pathologies through the analysis of the laboratory and physical exam results. Clinical problems such as polycythemia, sedentarism, hypercholesterolemia, and hypertriglyceridemia are prevalent among university workers. As for the pathologies, metabolic, hematologic are predominant. Ophthalmologic and musculoskeletal pathologies comprise significant percentages (38.6 %, 19%), respectively.
Alcohol and tobacco consumption presented minimal values of 8% and 3%. Finally, the occupational analysis revealed that nearly 84% of the employees are apt to work in their current job position, more than 16% are apt to work but with specific restrictions.
Consequently, there are no cases of employees that are classified as unfit to work.
The graphics presented below exhibit some biometric traits and associations with laboratory tests and the physical examination results.
55
Figure 9: Occurrence of abnormal laboratory test results by occupation PW-Total: Total number of individuals that perform physical work.
Figure 10: Occurrence of abnormal laboratory test results by gender Female- Total: Total number of females
56
Figure 11: Occurrence of abnormal laboratory test results by age
>41 Total: Total number of individuals that are 41 years old and higher.
Figure 12: Occurrence of abnormal laboratory test results by occupational diagnosis FL-Total: Total number of individuals that are fit with limitations.
57
Figures 9, 10, 11, and 12 show the abnormal results detected in the laboratory exams performed on the blood and urine samples. Each graphic dashed lines serve as a reference point to the specific category total number. For example, in Figure 9, the dashed line represents the total number of individuals that perform physical work (48). At the same time, the x-axis displays the total number of individuals that carry out mental work (more than 350), of which approximately 200 have a high hematocrit count.
The high count of hematocrit is a preponderant abnormal result in the laboratory exams for the whole workforce, followed by high cholesterol and triglycerides. Regarding the type of occupation displayed in Figure 9, the three abnormalities (hematocrit, cholesterol, and triglycerides) are prevalent for mental and physical work. Simultaneously, high uric acid, GOT, and GPT affect more mental work individuals than physical work.
Figure 10 shows a considerable high hematocrit count, cholesterol, and triglycerides in the male population compared with the females. Regarding the age in Figure 11, there is no clear distinction between the laboratory exam abnormalities and the age range. In the case of occupational diagnosis presented in Figure 12, abnormal parameters identified in the laboratory alone do not affect the fitness to work classification.
Figures 13, 14, and 15 reveal the problems determined during the physical exam and biometric traits such as gender, age, type of occupation, and occupational diagnosis.
58
Figure 13: Occurrence of problems in the eyes by gender, age, occupation, and occupational diagnosis
Female: Total number of females
<41: Total number of individuals that are 41 years old and higher PW: Total number of individuals that perform physical work FwL: Total number of individuals that are fit with limitations
Eye problems, which were the most encountered issue among the workforce, are depicted in Figure 13. Concerning gender, there is a higher proportion of males with eye problems than females. There is a minor difference regarding age. Employees in the age range of 15 to 41 years present more eye problems than their older colleagues. In terms of occupation, a considerable proportion of employees perform mental work and suffer a decrease in visual acuity.
Figures 14 and 15 show a summary of the physical exam results divided into the upper and lower body. Although eye affections were higher than physical problems, the upper and lower body issues are essential for recognizing work-related musculoskeletal disorders.
59
Figure 14:Occurence of problems in the upper body by gender, age, occupation and occupational diagnosis
Upper limb problems portrayed in Figure 14 show a significant proportion of back problems in all the categories. Female workers suffer more upper body problems (back, neck, wrists) than males. More individuals in the age range of 15 to 41 years old suffer upper body problems than workers older than 41. Mental work presents a higher count of cases in contrast to physical work. Back and upper limb problems are more significant in employees that are categorized as fit but with limitations.
60
Figure 15: Occurrence of problems in the lower body by gender, age, occupation, and occupational diagnosis
Problems in the lower body depicted in Figure 15 occurred with less frequency than eyes and upper body issues. Problems in knees are predominant in males, individuals older than 41 years old, mental work, and fit with limitations. On the type of occupation, legs, feet, veins, and hip problems dominate for employees in mental work positions.
Moreover, there is a larger number of individuals with problems in the lower parts of the body that are categorized as fit with limitations than fit to work.
61
Figure 16:Visualization of how biometric characteristics determine Fitness to work
The scheme presented in Figure 16 shows a summary of several biometric characteristics and the study results. It compares biometric features (gender, BMI classification, results obtained from the blood and urine tests) using an alluvial diagram[130]. It eases the visualization of how these characteristics are related to each other and the flow towards the final categorization on whether an employee is apt or apt with limitations to work.
The graphic shows the distribution of workers in each of the categories, sorted by size. A large proportion of individuals perform mental work. From this category, a significant portion is overweight or obese. Concerning the laboratory results, a high percentage of the individuals present abnormal results in the blood and urine exams, while the proportion is diminished in the physical examination problems. Only a small portion of the workforce is fit to work, but with some limitations. There are no cases where the employee is not fit to work.
62 2.2.4.2 Inferential Statistics
In addition to the descriptive statistics shown in the previous section. Inferential statistics were used for hypotheses testing. These tests were performed considering that the majority of the acquired primary data is categorical.
The tests presented below were executed using IBM SPSS 20 statistic software package.
1. Cronbach’s alpha coefficient reliability test Purpose
The reliability of a measurement resides on how consistent it measures a concept.
Cronbach’s alpha coefficient test constitutes a method to measure the internal consistency strength or the reliability of the items within a group or a concept.
Alpha (α) coefficient values range from zero to one[131][132]. Values near zero indicate that the items measured are independent of each other, whereas values that reach one demonstrate consistency in the items measuring the concept [133], [134].
In this case, the studied concept is the occupational diagnosis: whether an employee is fit to work, fit with limitations, or unfit to work.
Data Used
The items used for calculating the α coefficient value were the biometric screening outcomes listed below:
Laboratory exam results
Physical exam results
Pathologies
Clinical problems
The coding of the variables in SPSS ranged from zero to eight. Zero indicates the absence of problems in the laboratory exams, physical exams, pathologies, and clinical problems.
From one to eight, it indicates the number of abnormal results in laboratory exams, physical exams, encountered pathologies, and clinical problems detected in the worker.
63 Results
The α value calculated for the four items is 0.797, which is considered a reasonably high value[135][136].
Appendix II shows the outputs obtained in SPSS software.
2. Chi-Square Test Purpose
Chi-Square (χ2) test for independence was performed on the acquired biometric characteristics. It explores the relationships between the acquired characteristics and establishes if two categorical variables are statistically independent[136].
Chi-square tests are used to carry out hypothesis testing in nominal and ordinal data, in which the hypotheses are:
Null Hypothesis (H0): Two variables tested are statistically independent.
Alternative Hypothesis (H1): Two variables tested are statistically related.
This test contrasts the observed frequencies (number of observations acquired in each category) and expected frequencies (number of expected observations in each category considering the null hypothesis as true). Chi-square is calculated using the following equation[137]:
χ2 = ∑(𝑂 − 𝐸)2 𝐸 O = number of observed frequencies
E= number of expected frequencies
The calculated value is compared with the critical chi-square distribution value that depends on the degrees of freedom and the alpha level. Hence, if the critical value is less than the calculated chi-square value, the null hypothesis is rejected. The alternative hypothesis is accepted, concluding that there is a relationship between the two studied variables.
64 Data Used
The variables enumerated in Table 5 were used for determining what biometric characteristics are related and which ones are independent of the variables: type of occupation and occupational diagnosis. Table 5 also includes the categories of each variable and the coding used in SPSS.
Variables Categories Coding in
SPSS
Type of Occupation Physical Work 1
Mental Work 0
Occupational Diagnosis Fit 1
Fit with Limitations 0
Gender Female 0
Laboratory Results* Normal 0
Abnormal Results 1
Physical exam results* No problems 0
Problems 1
Table 5: Variables and categories used for chi-square analyses
* collapsed variables
The variables BMI classification, blood and urine laboratory results, and physical exam results were collapsed in order to meet the chi-square test assumption of the minimum expected cell frequency. Any cell should be five or more, and in tables 2x2 (two categories for each variable), the expected frequency should be at least ten. Moreover, SPSS shows at the end of the analysis in footnote a, if this assumption was violated, see Appendix II [136].
The majority of the analyses, besides the one including BMI classification, were done for tables 2x2. In this case, an adjustment to the chi-square called “Yates’ continuity correction” calculated in SPSS was used to determine if there is an association between the two studied variables. The continuity correction value is calculated to compensate for the overestimate of the chi-square value obtained in tables 2x2[136][138].
65 Results
Table 6 and Table 7 display chi-square test results and enumerate the biometric characteristics that have a significant relationship and which are not related to the variables: type of occupation and occupational diagnosis.
Characteristic Chi-Square tests of Independence Result
Gender
Table 6: Results from Chi-Square tests between collected biometric characteristics and type of occupation
66
Characteristic Chi-Square tests of Independence Result
Gender
Table 7: Results from Chi-Square tests between collected biometric characteristics and Occupational Diagnosis
67
The outputs retrieved from SPSS, detailing contingency tables and chi-square test results, are reported in Appendix II.
3. Logistic Regression Purpose
Logistic regression is used to model the relation between a discrete outcome and a group of continuous and categorical variables or a mix. It estimates the probability of an event occurrence for a randomly picked observation than the probability of the event not occurring. It also predicts the effect of several variables on a dichotomous dependent response and classifies the observations via the probability estimation of an observation falling into a particular category[139][140].
The estimated regression equation is[141]:
𝑌̂𝑖 = estimated probability ith case is in one of the categories (i= 1,…., n) u= linear regression equation:
A= constant
Bj = coefficients
Xj = predictors for k predictors
j= 1,…., k
Logistic regression was used to determine which biometric characteristics predict the outcome (occupational diagnosis), how these characteristics affect the occupational diagnosis, and if a particular characteristic or characteristics increase or decrease the probability of the outcome or if there is no effect on it.
68 Data Used
The dichotomous categorical outcome was: Occupational diagnosis. The predictors used were a mix of categorical and continuous variables. The categorical variables are:
Gender
Age
Alcohol consumption
Tobacco consumption
Laboratory results
Physical exam results
Pathologies
Clinical problems
Type of occupation The continuous variables are:
Height measured in meters
Weight measured in kilograms
The outcome and categorical variables codification facilitate the interpretation of the final results. Hence, a convenient way of coding is to consider the characteristic of interest, assigning a higher code (1) to the category of the variables most associated with the outcome category: the occupational diagnosis: Fit with limitations.
As shown in Table 8, the coding for the categories in occupational diagnosis is one for Fit with limitations and zero for Fit. Likewise, the categories of the variables associated with Fit with limitations such as alcohol, tobacco consumption, physical exam problems, abnormal laboratory results, and others are also coded one. In contrast, the categories associated to Fit are coded zero.
69
Variables Categories Coding in SPSS
Occupational Diagnosis Fit 0
Fit with Limitations 1
Gender Female 0
Male 1
Age 15-41 years 0
> 41 years 1
Alcohol Consumption
Yes 1
No 0
Tobacco Consumption
Yes 1
No 0
Laboratory Results Normal 0
Abnormal Results 1
Physical exam results No problems 0
Problems 1
Pathologies Yes 1
No 0
Clinical Problems Yes 1
No 0
Type of Occupation Physical Work 1
Mental Work 0
Table 8: Categorical variables used in Logistic Regression Analysis
In order to test multicollinearity, collinearity diagnostics were performed using Collinearity Statistics in SPSS. Table 9 indicates the tolerance values for the variables used in the logistic regression analysis. Low tolerance values (less than 0.1) show high correlations between that variable and other variables in the model[136]. As displayed in Table 9, none of the variables tolerance values are below 0.1, which indicates that the studied variables do not present high correlations with each other or multicollinearity.
70
Variables Tolerance
Gender 0.663
Age 0.926
Alcohol Consumption 0.884 Tobacco Consumption 0.883 Laboratory Results 0.801 Physical exam results 0.899
Pathologies 0.75
Clinical Problems 0.665 Type of Occupation 0.872
Height 0.264
Weight 0.146
Table 9: Collinearity Statistics variables Logistic Regression
Logistic regression was carried out using the SPSS procedure denominated Binary Logistic because the outcome studied (occupational diagnosis) has two categories (Fit and Fit with limitations). The Forced Entry Method was employed, which consists of testing all the predictors in one block to examine their predictive ability while controlling the effects of other predictors in the model.
Results
The model including all the predictors was statistically significant χ 2= 46.694, degree of freedom (df) = 11, p<0.001, N=409. Additionally, the Hosmer-Lemeshow Goodness of Fit indicates a good fit with a significance value of 0.499 (greater than 0.05), as shown in Table 10, which supports the model.
Chi-Square df Sig 7.36 8 0.499
Table 10: Hosmer and Lemeshow Test Results
Table 11 enumerates the different variables used in the model and their contribution as predictor variables. Column B displays the B coefficients used in the estimated regression equation to calculate the probability of a case falling in a specific outcome category.
Negative or positive B values indicate the direction of the relationship between the outcome and the factors.
71
Column P indicates the factors that are significant to the model. The two main factors influencing Fit with limitations are gender p=0,000 and physical exam results p=0.000.
The odds ratio for gender is 0.295 and for physical exam results is 3.879. These values indicate that being a male employee decreases by a factor of 0.295 the odds of being diagnosed as Fit with limitations. While employees having problems in the physical exams were 3.879 times more likely to be diagnosed as Fit with limitations than those without problems in the physical exam, controlling all of the other factors in the model.
The rest of the biometric characteristics do not significantly contribute to the variable
Table 11: Logistic Regression Predicting the Likelihood of being diagnosed as Fit with limitations
72
DISCUSSION
In this section, the results obtained through the experimental methods (preliminary research and case study) are first examined and explained. Additionally, the hypotheses previously formulated are discussed and contrasted with the experimental results to validate or reject the research hypotheses.
Preliminary Research
The preliminary research objective was to acquire a baseline regarding wellness programs aspects, especially biometric screenings. The survey administered to the employees gathered their perceptions, opinions, and suggestions to make wellness initiatives, including biometric screenings, as valuable for the company as they are for the employee.
The survey participation rate was low. The answers of 89 respondents were used to present the results. Even though the availability of wellness programs and initiatives is continuously increasing in corporations[142], workers’ participation is low[143][144].
For this particular survey, more advertisements and a more extended period to collect data could have given better results. In addition, targeting groups such as young[145] [146]or health-conscious employees[70] can increase participation rates.
This survey helped the company know about the employees’ perceptions of health programs and their primary needs regarding its health initiatives. For example, a high
This survey helped the company know about the employees’ perceptions of health programs and their primary needs regarding its health initiatives. For example, a high